Stochastic seismic lateral deformation of a multi-story subway station structure based on the probability density evolution method

Effective utilization of urban underground space has seen multi-story subway station structures become increasingly common. Two significant issues need to be solved in seismic evaluation of this kind of station structure. One is to set the appropriate seismic performance evaluation index; another is to choose a comprehensive method that can evaluate the structure performance with regard to the index. In this paper, a numerical model was established and validated by previously conducted shaking-table tests. Then stochastic seismic analyses were performed on the model using the probability density evolution method (PDEM). Inter-story drift and column drift ratio, which can reflect the lateral deformation of station structures induced by earthquake, were selected as seismic performance evaluation indexes and studied from the stochastic view. The stochastic analysis showed that the probability density distribution of the inter-story drift extremum of the middle-stories of the structure is larger than that of the top story and bottom story, similar to the lateral deformation mode of a ground frame structure with strong columns and weak beams. In addition, the probability density distribution of the column drift ratio extremum in each story is quite similar, and the probability density distribution of the column drift ratio extremum is larger than the inter-story drift extremum of the corresponding story. Compared with the inter-story drift, the column drift ratio is more suitable for evaluating the seismic performance of the multi-story subway station structure, because it reflects the influence of longitudinal beams with different height at the two ends of column on the lateral deformation of columns, which are key components and determine the safety of the structure to a great extent. By comparing the results of the PDEM with the shaking-table test, we found that stochastic analysis method can further obtain the probability of each numerical value of seismic performance evaluation indexes, thus making the evaluation results more comprehensive.